Synthesis and Reactions of 3d Metal Complexes with the Bulky

Article pubs.acs.org/IC

Synthesis and Reactions of 3d Metal Complexes with the Bulky Alkoxide Ligand [OCtBu2Ph] James A. Bellow,†,¶ Maryam Yousif,†,¶ Dong Fang,†,‡ Eric G. Kratz,† G. Andrés Cisneros,*,† and Stanislav Groysman*,† †

Department of Chemistry, Wayne State University, Detroit, Michigan 48202, United States Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin 53706, United States

‡

S Supporting Information *

ABSTRACT: Treatment of NiCl2(dme) and NiBr2(dme) (dme = dimethoxyethane) with 2 equiv of LiOR (OR = OCtBu2Ph) forms the distorted trigonal planar complexes [NiLiX(OR)2(THF)2] (THF = tetrahydrofuran) 5 (X = Cl) and 6 (X = Br). The reaction of CuX2 (X = Cl, Br) with 2 equiv of LiOR affords the Cu(I) product Cu4(OR)4 (7). The same product can be obtained using the Cu(I) starting material CuCl. NMR studies indicated that the reduction of Cu(II) to Cu(I) is accompanied by the oxidation of the alkoxide RO− to form the alkoxy radical RO•, which subsequently forms tert-butyl phenyl ketone by β-scission. Treatment of compounds 1−4 ([M2Li2Cl2(OR)4], M = Cr− Co) with thallium hexafluorophosphate allowed the isolation of the distorted tetrahedral complexes of the form M(OR)2(THF)2 for M = Mn (8), Fe (9), and Co (10). Cyclic voltammetry performed on compounds 8−10 demonstrated irreversible oxidations for all complexes, with the iron complex 9 being the most reducing. Complex 9 shows a reactivity toward PhIO and Ph3SbS to form the corresponding dinuclear iron(III) complexes Fe2(O)(OR)4(THF)2 (11) and Fe2(S)(OR)4(THF)2 (12), respectively. X-ray structural studies were performed, showing that the Fe−O−Fe angle for complex 11 is 176.4(1)° and that the Fe−S−Fe angle for complex 12 is 164.83(3)°.

■

INTRODUCTION

alkoxides (and related aryloxides) are capable of supporting Cu(III) complexes. We seek low-coordinate mononuclear complexes that, upon reaction with an organic azide, will form reactive nitrene functionality. We surmised that alkoxide ligation at a middle-tolate 3d metal would lead to highly reactive metal−nitrene as (i) alkoxides are weak σ-donors, which should render metal− nitrene electron-deficient and (ii) at the same time, alkoxides are capable of π-donation, which is expected to interfere with πbonding from the nitrene functionality. As we are interested in π-basic alkoxides, we decided to pursue a nonfluorinated bulky alkoxide ligand. Our ligand of choice was [OCtBu2Ph].11 The steric bulk of [OCtBu2Ph] is comparable to the bulkiest alkoxide, [OCtBu3], previously utilized by Wolczanski and coworkers7 and Power and co-workers.8 By replacing one of the t Bu groups by a Ph group we hoped to improve its packing and crystallinity. Furthermore, the phenyl group is tunable, allowing the ability to add additional steric bulk if needed. In our previous publications, we reported that [OCtBu2Ph] (OR thereafter) reacts with MCl2 precursors (M = Cr, Mn, Fe, and Co) to form novel alkoxide clusters of [M2Li2Cl2(OR)4] composition featuring rare seesaw geometry at transition metal

Alkoxides and related phenoxides and siloxides are commonly used as ancillary ligands in coordination chemistry and catalysis.1−4 However, the majority of alkoxide complexes involve early transition metals, as alkoxides are π-donors and bind stronger to the more electron-deficient metals.5 Welldefined mid- and late-transition metal complexes of alkoxides are relatively rare due to the destabilizing interaction of the filled d orbitals on the metal with the p orbitals on the alkoxides.1 The ensuing basicity of the alkoxide lone pairs leads to the formation of polymetallic clusters via bridging.1 The proclivity of alkoxides to form clusters can be overcome by two different approaches: (1) making the alkoxide sterically hindering,6−9 or (2) decreasing the nucleophilicity of the alkoxide lone pairs.10 Demonstrating the first approach, Power and co-workers reported middle and late transition metal complexes stabilized by alkoxides featuring three large groups at the central carbon (OCR3, R = tBu, Cy, Ph).8 Pursuing the second approach, Doerrer and co-workers have recently described late transition metal complexes ligated by perfluorinated alkoxides.10 They have demonstrated that fluorination diminishes π-basicity, prevents bridging, and leads to the formation of stable monomeric low-coordinate late (Ni, Cu) metal centers. They also demonstrated that fluorinated © 2015 American Chemical Society

Received: December 11, 2014 Published: June 4, 2015 5624

DOI: 10.1021/acs.inorgchem.5b00795 Inorg. Chem. 2015, 54, 5624−5633

Article

Inorganic Chemistry Scheme 1. Synthesis of the Alkoxide Complexes 1−12 Reported in This Paper

centers (1−4, Scheme 1 below).11a For Fe, we have also demonstrated that [Fe2Li2Cl2(OR)4] can be transformed into a mononuclear complex Fe(OR)2(THF)2 (THF = tetrahydrofuran), which displays intriguing reactivity with an alkyl (adamantyl, Ad) azide, enabling its one-electron reductive coupling to form the metal-ligated hexazene species [AdNNNNNNAd]2−.11b Following these initial results, we became interested in the synthesis of other middle-to-late transition metal bis(alkoxide) complexes, specifically, those of Cr, Mn, Co, and Ni. Furthermore, the intriguing reactivity of the Fe(OR)2(THF)2 complex with azides prompted us to investigate the reactivity of Fe(OR)2(THF)2 and other alkoxide complexes in our study in related atom- and group-transfer reactions. Herein we report the coordination chemistry of [OCtBu2Ph] with 3d metals (M = Cr − Cu) and the reactions of the resulting metal complexes with [O]- and [S]-atom transfer reagents.

the three-coordinate intermediate [MLiCl(OR)2(THF)2], which was directly observed only for Fe.11a In contrast, the reaction of NiCl2(dme) (dme = dimethoxyethane) with 2 equiv of LiOR followed by recrystallization from hexanes forms stable orange [NiCl(OR)2Li(THF)2] (5, 65% yield). Similarly, the reaction of NiBr2(dme) with 2 equiv of LiOR forms red-orange [NiBr(OR)2Li(THF)2] (6, 63% yield). Complexes 5 and 6 demonstrate solution magnetic moments of 3.0 ± 0.2 and 3.1 ± 0.2 μB, respectively, indicating two unpaired electrons (expected spin-only moment is 2.8 μB). The structure of 6 is given in Figure 1, and the structure of isomorphous 5 is given in

■

RESULTS AND DISCUSSION Reactivity of MX2 (M = Cr − Ni) with LiOR: Formation of Seesaw Clusters [M2Li2Cl2(OR)4] Versus Heterodinuclear [NiLiX(OR)(THF)2] Complexes. Scheme 1 describes the formation of various [M(OR)2] complexes obtained in this study. Four distinctly different structural types were observed: (i) [M2Li2Cl2(OR)4] clusters in which the seesaw transition metal centers are ligated by two [OR] and two Cl ligands,11a (ii) heterodinuclear [MLiX(OR)2(THF)2] complexes displaying trigonal transition metal centers with two [OR] and one X ligands, (iii) mononuclear [M(OR)2(THF)2] complexes, and (iv) tetranuclear Cu4(OR)4 complex. As previously described, treatment of CrCl2, MnCl2, FeCl2, and CoCl2 with 2 equiv of LiOR in THF, followed by recrystallization from hexanes, affords [M2Li2Cl2(OR)4] clusters.11a The reactions were postulated to proceed through

Figure 1. Structure of [NiLiBr(OR)2(THF)2] (6), 50% ellipsoids. H atoms were omitted for clarity. Dashed lines indicate largely ionic interactions with lithium.

the Supporting Information. Both compounds display distorted trigonal planar geometry at the Ni centers (sum of angles at Ni 360.0°). Selected structural data for 5 and 6 are presented in Table 1. For comparison, we included the structural data for one of the seesaw clusters (3). 5625

DOI: 10.1021/acs.inorgchem.5b00795 Inorg. Chem. 2015, 54, 5624−5633

Article

Inorganic Chemistry Table 1. Selected Structural Data for the Compounds Reported in This Paper complex 3 5 6 7 8 9 10 11 12 a

M−ORterminal (Å)

M−ORbridging (Å)

OR−M−OR (deg)

1.748(1) 1.747(1)

1.91(1)a 1.871(1) 1.872(1) 1.85(1)d

174.0(2) 147.67(6)c 148.24(5)c 165(4)b 130.0(1) 138.7(1) 131.1(1) 108.6(1) 117(4)a

1.899(1) 1.838(1) 1.8491(1) 1.82(1) 1.80(1)b

THF−M−THF (deg)

M−X−M (deg)

M = X (or M−X−M) (Å)

176.4(1) 164.83(3)

1.8031(3) 2.186(4)a

86 83 88

Average of two bonds/angles. bAverage of four bonds/angles. cThe angle between terminal and bridging alkoxides. dAverage of eight bonds.

Reactivity of Cu(II) and Cu(I) Precursors with LiOR: Formation of Cu4(OR)4 Tetramer. We also investigated the reactivity of copper(II) precursors with LiOR. Treatment of an orange solution of CuBr2 in THF with 2 equiv of LiOR in ether forms immediately a dark red solution. The red color, however, instantaneously changes to amber, light blue, and eventually to colorless. Solvent removal, followed by recrystallization of the product from hexanes at −35 °C, leads to the formation of the colorless Cu(I) cluster [Cu4(OR)4] (7) isolated in 42% yield. The reaction of CuCl2 with LiOR under identical conditions forms the same product in 46% yield. [Cu4(OR)4] (7) can be also obtained from a copper(I) precursor, CuCl, in 51% yield. We characterized compound 7 by 1H and 13C NMR spectroscopy, X-ray crystallography, and elemental analysis. The 1H NMR spectrum of 7 contains one kind of OR resonance, consistent with the approximate D2d symmetry of the structure. The X-ray structure of compound 7 is given in Figure 2. Compound 7 is a tetramer of [CuOR] units featuring

Most remarkably, all of the previous Cu4(OR)4 compounds were obtained from copper(I) precursors, while in our case it is also obtained from various copper(II) starting materials. It is possible that at the first step, an analogous (to [NiLiBr(OR)2(THF)2] or [FeLiCl(OR)2(THF)2]) product [CuLiX(OR)2(THF)2] is formed (see below for its calculated structure). We also note that calculations (see below) indicate that its dimerization to form [Cu2Li2X2(OR)4] is unlikely. The copper center in [CuLiX(OR)2(THF)2] then undergoes reduction, and the alkoxide constitutes the only possible reducing agent in such transformation. Upon oxidation, it is expected to form an [RO] radical. The cyclic voltammetry (CV) of LiOR demonstrates an irreversible oxidation at the relatively accessible potential of 0.51 V versus FeCp2/FeCp2+ (Figure 3). We note that tertiary alkoxide is not commonly

Figure 3. CV of LiOR in THF, (0.1 M [NBu4](PF6), 25 °C, platinum working electrode, 100 mV/s scan rate).

observed as a reductant.1 Furthermore, we are unaware of any previous examples in which Cu(II) is reduced by an alkoxide. We also note that the inability of [OR] to support even Cu(II) centers stands in sharp contrast to the behavior of fluorinated alkoxides capable of supporting Cu(III).10 We followed the reaction of CuBr2 with 2 equiv of LiOR by UV−vis spectroscopy and NMR spectroscopy. The addition of 0.5 mL of a 2.86 M LiOR solution into an orange 0.286 mM solution of CuBr2 (2.5 mL in THF) immediately leads to a color change to colorless. A UV−vis scan performed ca. 10 s after the addition revealed the final product (Supporting Information, Figure S22). We tried to perform the reaction at 0 or at −78 °C. Still, the reaction was too fast at 0 °C, and no reaction took place at −78 °C. However, batch-wise addition of the solution of LiOR to the solution of CuBr2 (Figure 4) at room temperature (RT) provided clear evidence for the existence of an intermediate. The red trace in Figure 4

Figure 2. Structure of [Cu4(OR)4] (7), 50% ellipsoids. H atoms and THF solvent were omitted for clarity.

nearly linear Cu(I) centers each ligated by two alkoxides. Four Cu(I) centers form a plane with Cu···Cu distances ranging from 2.64 to 2.72 Å. The alkoxides coordinate below and above the Cu4 plane, leading to the overall “butterfly” structure. Such geometry can be also described in terms of two Cu2(OR)3 planes (O4Cu4Cu3O1O3 and O2Cu1Cu2O1O3) intersecting at 115° angle. Several related compounds exist in the literature, namely, Cu4(OtBu)4, Cu4(OAr)4 (Ar = C6H3Ph2), and Cu4(OCPh3)4.1,12 Of the above, Cu4(OtBu)4 and Cu4(OAr)4 are planar, while Cu4(OCPh3)4 has a similar (to compound 7) butterfly molecular geometry, with a wider angle (141°) between the Cu2(OR)3 planes. 5626

DOI: 10.1021/acs.inorgchem.5b00795 Inorg. Chem. 2015, 54, 5624−5633

Article

Inorganic Chemistry

Figure 4. UV−vis spectra following the batch-wise addition of LiOR to the solution of CuBr2. (inset) The spectrum of the Cu4(OR)4.

Figure 5. (A) 1H NMR spectrum of an aliquot of the crude reaction mixture displayed at 6.8 to 8.8 ppm. (B) 1H NMR spectrum of Cu4(OR)4 in the same region. (C) 1H NMR spectrum of tert-butyl phenyl ketone (2,2-dimethyl-1-phenylpropan-1-one). (D) 1H NMR spectrum of HOR.

represents the UV−vis spectrum of the 0.287 mM solution of CuBr2 in THF (2.5 mL) that is dominated by two signals at 415 and 355 nm. The addition of 0.1 mL of a 2.87 mM solution of LiOR (0.4 equiv) leads to the decline in the intensity of the absorption peaks attributed to CuBr2 and an appearance of a new signal at 660 nm. The spectrum of the final product of the reaction (Cu4(OR)4, see the black spectrum in the inset of Figure 4) is nearly featureless. Thus, the peak at 660 nm belongs to the short-lived intermediate of the reaction. The addition of another batch of LiOR (purple trace, next 0.4 equiv) causes further increase in the intensity of the signal at 660 nm, which is accompanied by a decrease in the intensity of the signals at 415 and 355 nm. The next addition (orange trace) leads to a further decrease in the spectrum of the starting

material, while the signal attributed to the intermediate persists. Further additions of the alkoxide (green and black traces, 2 equiv total of LiOR) lead to the disappearance of the 660 nm signal. To gain further insight into the reaction mechanism, we also analyzed the reaction by means of 1H NMR spectroscopy. UV− vis spectroscopy indicates that the reaction is complete within several minutes. The reaction of CuBr2 with 2 equiv of LiOR in tetrahydrofuran in the presence of internal standard trimethoxybenzene was allowed to stir at RT for 30 min, after which an aliquot was taken, and its 1H NMR spectrum was collected in C6D6. Figure 5A demonstrates the aromatic region of the spectrum; the full spectrum is given in Figure S6 in the Supporting Information. For comparison, Figure 5 also 5627

DOI: 10.1021/acs.inorgchem.5b00795 Inorg. Chem. 2015, 54, 5624−5633

Article

Inorganic Chemistry Scheme 2. Proposed Pathway to the Formation of [Cu4(OR)4], tert-Butyl Phenyl Ketone, and HOR

demonstrates NMR spectra of clean Cu4(OR)4 (B), tert-butyl phenyl ketone (C), and HOR (D) presented in the same region. Full spectra are given in the Supporting Information. Spectrum A demonstrates that the crude reaction mixture contains Cu4(OR)4. In addition, resonances attributed to tertbutyl phenyl ketone and HOR are observed as well. Using the internal standard (trimethoxybenzene), we quantified the overall amount of the “[OR]” or the OR derivatives detected (quantification was done based on the tert-butyl resonance). We observe formation of 0.012 mmol of Cu4(OR)4 (0.048 mmol of “Cu(OR)”), 0.026 mmol of tert-butyl phenyl ketone, and 0.022 mmol of HOR. Given the amount of LiOR used (0.156 mmol), we account for ∼62% of [OR]; it is consistent with observed yield of Cu4(OR)4 (60%). Furthermore, these data also suggest that for each equivalent of Cu(OR), we are forming ∼0.5 equiv of tert-butyl phenyl ketone and HOR each. Combined with the UV−vis data presented in Figure 4, the detection and quantification of various reaction products provides a key to the understanding of the reaction mechanism (Scheme 2). We propose that [CuLiBr(OR)2(THF)2] forms first, as observed or proposed for Cr−Ni. Next, the Cu−OR bond undergoes a homolytic bond cleavage to form “CuI(OR)”, which tetramerizes to form Cu4(OR)4. The remaining alkoxy radical undergoes β-scission, which has been previously observed for alkoxy radicals.13 The β-scission produces tertbutyl phenyl ketone and tert-butyl radical, which may decompose to give isobutylene and hydrogen atom that couples with the alkoxy radical to give HOR. The formation of HOR is also possible via the H atom abstraction by the alkoxy radical from the solvent.13 We did not observe isobutylene: the reaction takes place in THF/ether mixture, and the solvent/volatiles needed to be removed prior to dissolving the reaction mixture in C6D6. Synthesis and Characterization of Mononuclear Bis(alkoxide) complexes M(OR)2(THF)2. At the next step, we turned to the synthesis of neutral mononuclear species lacking complexed LiX. We postulated that complexed halide anions can be abstracted by the addition of Tl+, which is expected to form insoluble TlX. Treatment of [M2Li2X2(OR)4] (2−4)

complexes with TlPF 6 in THF affords colorless Mn(OR)2(THF)2 (8, 60%), colorless Fe(OR)2(THF)2 (9),11b and violet Co(OR)2(THF)2 (10, 49%). We note that compounds 8−10 are related to the previously described M(OR′)2(THF)2 (OR′ = OCPh3, M = Mn, Fe, Co)8e,f and the more recently synthesized Mn(OAr)2(THF)2 complexes.14 Surprisingly, no reaction with TlPF6 was observed for the heterodinuclear Ni complexes 5 and 6. Workup of the reaction mixture in both cases led to the reisolation of the starting materials. Treatment of the Cr-containing seesaw cluster 1 with TlPF6 failed to produce the corresponding Cr(OR)2(THF)2 species. Compounds 8−10 were characterized by IR spectroscopy, 1 H NMR spectroscopy, solution magnetic measurements, UV− vis spectroscopy, CV, and X-ray crystallography. Compounds 8−10 demonstrate μeff values of 5.5 ± 0.3 μB for Mn (8), 4.7 ± 0.3 μB for Fe (9), and 3.6 ± 0.3 μB for Co (10) at RT (measured by Evans method using average of three measurements). The corresponding calculated spin-only values for the high-spin Mn(II), Fe(II), and Co(II) centers are 5.9, 4.9, and 3.9 μB, respectively. Thus, the observed magnetic moments of distorted tetrahedral complexes are close to spin-only magnetic moments. We note that two-coordinate bis(aryloxide) Fe and Co complexes demonstrate significantly higher μeff values due to a considerable orbital contribution.15,16 The IR spectra of 8− 10 are given in Supporting Information, Figure S12. As anticipated, compounds 8−10 give rise to almost identical IR spectra with prominent C−O stretches around 1100 cm−1. UV−vis spectra are given in the Supporting Information. Colorless 8 and 9 are nearly featureless in the observed region. Pale violet 10 (Co(OR)2(THF)2) gives rise to several weak transitions in the visible region. The most significant peak at 734 nm exhibits an extinction coefficient of 175 M−1 cm−1. CVs of compounds 8−10 are given in Figure 6. CVs were obtained in THF, using [NBu4](PF6) as the electrolyte, and are reported versus ferrocene/ferrocenium couple. Only irreversible oxidations are observed, peaking at 0.45 V for Fe(OR)2(THF)2 (onset at ∼0 V), 1.27 V for Co(OR)2(THF)2 (onset at ∼0.5 V), and 1.43 V for Mn(OR)2(THF)2 (onset at 5628

DOI: 10.1021/acs.inorgchem.5b00795 Inorg. Chem. 2015, 54, 5624−5633

Article

Inorganic Chemistry

are summarized in Table 1. Similarly to the previously described 9, compounds 8 and 10 are distorted tetrahedral complexes featuring wide RO−M−OR angles (130° for 8, 131° for 10), and narrow THF−M−THF angles (86° for 8, 88° for 10). Related distorted tetrahedral Mn(OAr)2(THF)2 and Fe(OCPh3)2(THF)2 were previously reported, both exhibiting RO−M−OR angles of 140°.14,8f Atom-Transfer Chemistry of Complexes 8−10. The electrochemical studies indicate that Fe(OR)2(THF)2 (9) is more reducing than Mn(OR)2(THF)2 and Co(OR)2(THF)2 complexes 8 and 10. The reactivity of 8−10 in group-transfer chemistry correlates with these observations. We have previously demonstrated that 9 reductively couples adamantyl azide to form Fe(III)-bound hexazene.11b In contrast, no reaction with adamantyl azide was observed for the Mn and Co complexes 8 and 10. Similarly, 9 undergoes a reaction with an oxo-transfer reagent (PhIO) and a sulfur-atom transfer reagent (Ph3SbS)18 to furnish Fe2(O)(OR)4(THF)2 (11) and Fe2(S)(OR)4(THF)2 (12), respectively (Scheme 1). Treatment of 8 and 10 with the oxo- and sulfur atom-transfer reagents under similar reaction conditions failed to produce the oxidized products. The structures of 11 and 12 are given in Figure 9. Both complexes feature distorted tetrahedral Fe(III) centers ligated by two terminal alkoxides, one THF ligand, and a bridging oxo (11) or sulfido (12) ligand. Thus, similarly to the previously reported hexazene complex [Fe 2 III (AdNNNNNNAd)(OR)4],11b the Fe centers retain the bis(alkoxide) ligation upon chemical transformation. Complex 11 displays crystallographic C2 symmetry, with only half the molecule occupying the asymmetric unit. The Fe−O−Fe angle is 176.4(1)°, and the Fe−(μ2-O) distance is 1.8031(3) Å. The Fe−S−Fe angle in the non-C2-symmetric 12 is 164.8°, and the Fe−S bond distances are 2.1821(5) and 2.1901(5) Å. While Fe2(μ2-O) compounds are common in the literature, compound 12 is one of the few known di-Fe(III) complexes in which the Fe centers are bridged by a single sulfide.19 The IR spectra of compounds 11 and 12 in the 600−1700 cm−1 region are presented in Figure 10 below. The overlay of the spectra demonstrates a significant structural similarity, with the exception of the strong peak at 795 cm−1 that is observed only for compound 11 and is therefore assigned to the Fe−oxo stretch. The Fe−sulfido stretch is expected to occur in the far-IR region (